洪涝灾害条件下疏散交通生成预测方法

Method for predicting evacuation traffic generation under floods condition

为有助于有关部门更准确预测洪涝灾害受灾民众的疏散量,结合非集计数据和集计数据的优点,提出分区集计数据的概念,设计了受灾区域分区方法,并通过意向偏好(SP)调查法对我国居民在洪涝条件下疏散交通需求数据进行调查。在此基础上,引入BP神经网络建立基于分区集计数据的疏散交通生成预测模型。利用调查数据进行实证分析发现,所设计方法取得了较好的预测效果,鲁棒性较好,平均相对预测误差仅为1.8%,其预测效果明显优于现有的非集计和整集计模型。

To be helpful for related departments to predict the evacuation work load more accurately under floods condition,combining the advantages of aggregate data and disaggregate data,a concept of aggregate data of zone was proposed. A disaster-affected area partition method was designed,and the evacuation traffic demand data were obtained from residents using SP investigation method. A model was built for predicting evacuation traffic generation by using BP networks theory. The test data were obtained from SP investigation,the experimental results indicate that the performance of proposed method is good,which is better than that of the two existing models.